By STEPHEN HARDY
With service providers placing a premium on lowering costs and increasing their ability to quickly provision new services and customers, tunable lasers have become a hot technology. The first wave of devices, generally narrowly tunable, used standard distributed-feedback (DFB) and distributed Bragg reflector (DBR) approaches, with the occasional sampled grating technology mixed in to meet a need for wider tuning range. A new wave of devices has arrived this year and highlights how developers have turned to novel approaches to avoid trading off power for tunability.
The large number of new devices promises plenty of choices for systems designers. Recently announced offerings that diverge from the mainstream include the MetroFlex from Bandwidth9 Inc., a "programmable ITU laser" from Blue Sky Research, Iolon Inc.'s Apollo, the PowerTune from New Focus Inc., the ML-20-3062-1 from Nortel Networks/Coretek, and the PowerSweep 2000 from Princeton Optronics.
While their initial applications re main primarily sparing and inventory management, these lasers also attempt to meet future requirements for in-system tunability to support rapid service provision, flexible add/drop, and wavelength conversion, among other advanced functions.
"I don't think we could even begin to justify a business case if the application was only sparing," says Tom Dudley, vice president of business development at Nortel/Coretek. "I think the real application is in all-optical networks."
Device manufacturers appear to agree that their offerings must tune at least across the C-band and provide 20 mW of power to meet the full spectrum of anticipated uses. Bandwidth9, Blue Sky, and Nortel/Coretek are already discussing L-band coverage. Blue Sky promises to hit this milestone with its second release, slated for next year, while Nortel/Coretek expects its device to provide C- and L-band coverage when it achieves general availability this October. Bandwidth9 touts L-band tuning now.
Dudley reports that the 20-mW power requirement is not universal; a significant percentage need no more that 10 mW, he reports. Nortel/Coretek will provide versions of its device at both power levels. In this vein, Bandwidth9 offers 500 microwatts continuous wave for the metro applications it intends to reach with the MetroFlex.
Some degree of commonality has emerged among the new offerings. For example, several manufacturers have turned to external-cavity designs, including Blue Sky, Iolon, New Focus, and Princeton Optronics. (For information on the design and operation of external-cavity devices, see two articles in the March 2001 issue of Lightwave: "External-cavity diode lasers for ultra-dense WDM networks," page 130, and "Tunable-laser technologies vs. optical-networking requirements," page 136.)
Proponents of external-cavity devices claim their approach provides very narrow linewidths and high-frequency stability-particularly more immunity to temperature sensitivity than DFB's provide. Naysayers claim external-cavity designs are complex, expensive, and potentially more susceptible to the effects of shock and other environmental influences than more typical approaches.
These critics might point to the use of micro-electromechanical systems (MEMS) technology-the second major trend within the newest generation of tunable-laser devices-to bolster their point about complexity. Yet Iolon and Princeton Optronics, which use MEMS as one of the standard tuning components within their external-cavity de vices, express confidence that their approaches do not add enough complexity to compromise reliability and economics. Iolon in particular has touted its use of deep reactive ion etching techniques to fabricate the MEMS actuator as a path toward reliability and efficiency.
Even companies that don't base their designs on an external-cavity architecture have expressed enough faith in MEMS devices to use them as tuning agents, particularly Bandwidth9 and Nortel/Coretek. The latter's approach is particularly unique; according to Dudley, the MEMS device sits on the top mirror of the company's vertical-cavity surface-emitting laser (VCSEL). The MEMS acts as a lens, he says.
The use of VCSELs represents a third trend among the recent tunable-laser offerings. Besides Nortel/Coretek, Bandwidth9 and Princeton Optronics also use VCSELs, although in the latter case, Princeton uses a 980-nm VCSEL as a pump, not as the primary gain medium. VCSELs have previously resisted use within the C-band and in applications that require high power. Nortel/Coretek in particular has insisted that VCSEL architectures, in combination with optical pumping, can meet power requirements of 20 mW and beyond, as well as provide C- and L-band operation. In fact, the company claims that its VCSEL-based approach achieves the linewidth, relative intensity noise (RIN), and SMSR figures necessary for 40-Gbit/sec applications. Several customers are currently sampling such a device, Dudley claims.
The vendors mentioned above illustrate how many companies believe that a new path provides the best approach to both high power and wide tunability. However, not everyone agrees that more conventional approaches have no future. For example, Agere Systems has introduced the C92, a 10-Gbit/sec tunable transmitter that will tune across 20 channels spaced 50-GHz apart.
The device uses the company's electro-absorptive modulated laser (EML) technology and provides a laser, wavelength locker, and modulator in the same package. Meanwhile, Agility Communications says its DBR-based architecture will support C-band tuning, as well.
With most of the new wave of devices still in the sampling stage-or earlier in their development-the viability of these new approaches to tunability still awaits the judgment of systems designers. As consensus forms, equipment developers will learn which options hold the most promise for future system requirements.